ebook img

The Assessment Checklist for Adolescents – ACA - childpsych.org.uk PDF

26 Pages·2013·0.25 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview The Assessment Checklist for Adolescents – ACA - childpsych.org.uk

The Assessment Checklist for Adolescents - ACA PRE-PRINT MANUSCRIPT Tarren-Sweeney, M. (2013). The Assessment Checklist for Adolescents – ACA: A scale for measuring the mental health of young people in foster, kinship, residential and adoptive care. Children and Youth Services Review, 35, 384-393. Abstract The Assessment Checklist for Adolescents – ACA is a 105-item carer-report mental health rating scale, measuring behaviours, emotional states, traits, and manners of relating to others, as manifested by adolescents (ages 12 to 17) residing in various types of alternate care, as well as those adopted from care. The ACA was designed for population and clinical research with these young people, and for use as a clinical assessment measure. The ACA‟s content was largely derived from the Assessment Checklist for Children (ACC). Fifteen ACC items were modified to better reflect adolescent difficulties, and 25 additional items were derived using a combination of inductive and deductive strategies. Item and factor analyses were carried out on scores from a 136-item research instrument, obtained for 230 young people residing in long-term care (as part of the NSW Children in Care study). These data were supplemented by ACC scores obtained for 142 adolescents residing in treatment foster care in Ontario, Canada. A robust 7-factor model was identified among a core of 73 clinical item scores, accounting for 51% of score variance. Four of the factors replicate ACC clinical scales (non-reciprocal interpersonal behaviour; sexual behaviour problems; food maintenance behaviour; and suicide discourse), and three are unique to the ACA (social instability / behavioural dysregulation; emotional dysregulation / distorted social cognition; and dissociation / trauma symptoms). The ACA also contains two empirically- derived low self-esteem scales (low confidence; negative self-image) that are shared with the ACC. Initial data indicate that the ACA has good content, construct and criterion-related validity, as well as high internal reliability. Keywords Developmental psychopathology; attachment and trauma difficulties; foster care; adoption; mental health assessment; Assessment Checklist for Children 1 The Assessment Checklist for Adolescents - ACA 1. Introduction 1 The present paper describes the development of the Assessment Checklist for Adolescents (ACA), which is an adolescent version of the Assessment Checklist for Children (ACC). The ACC is a 120-item carer-report psychiatric rating instrument, that measures behaviours, emotional states, traits, and manners of relating to others, as manifested by children in care, and related populations (including children adopted from care) (Tarren-Sweeney, 2007). It was designed to measure a range of mental health difficulties observed among children in care that are not adequately measured by standard rating instruments, such as the Child Behavior Checklist (CBCL), the Strengths and Difficulties Questionnaire (SDQ) and the Conners scales. These consist of a number of attachment-related difficulties (indiscriminate, non-reciprocal and pseudomature types), insecure relating, trauma-related anxiety, abnormal responses to pain, overeating and related food maintenance behaviours, sexual behaviour problems, self-injury and suicidal behaviours and discourse. The ACC was developed a decade ago for use in the Children in Care Study (CICS), a prospective epidemiological study of the mental health of children in long-term foster and kinship care, in New South Wales (NSW), Australia (Tarren-Sweeney & Hazell, 2006). Aside from investigating the nature of the mental health of children in care, the CICS located evidence that pre-care social adversity factors (notably the extent of children‟s exposure to pre-care maltreatment) were stronger predictors of the mental health difficulties measured by the ACC, than were children‟s experiences in care. Since then, the ACC has been employed in over 20 studies in Europe, North America and Australasia, and is increasingly used as a clinical assessment tool by specialised mental health services for children and young people in care – particularly in Britain and Australia (Chambers, Saunders, New, Williams, & Stachurska, 2010; DeJong, 2010). Prior to commencing the CICS, the ACC‟s content was derived for a wide age range (4 to 17 years), with a view to: 1. employing a single set of items for baseline and follow-up studies; and 2. comparing the factor structure of this common set of items at different developmental stages. This plan was subsequently revised following examinations of what the ACC appears to measure among children and young people at different ages, and with different developmental pathways. A developmental perspective suggests that clinical phenomena are likely to evolve in complex ways in the context of both age-sensitive developmental stages, and maladaptive developmental trajectories. For this reason it was thought that, prior to embarking on a 7- to 9-year follow-up of 1 Abbreviations: CBCL Child Behavior Checklist CICS Children in Care Study NSW New South Wales, Australia ROC Receiver Operating Characteristics SDQ Strengths and Difficulties Questionnaire TFC Treatment Foster Care 2 The Assessment Checklist for Adolescents - ACA the CICS cohort (who would then be aged 11 to 18 years), there should be closer investigation of the validity and comprehensiveness of the ACC item pool for measuring those mental health symptoms manifested by adolescents in care that are not adequately measured by the CBCL, SDQ, Conners, etc. This preliminary investigation located a number of items that appear unsuitable for an adolescent population; suggested a need to modify some ACC items for an adolescent population; and suggested that some clinical phenomena manifested by young people in care are not adequately captured by the ACC. These findings prompted the formal development of an adolescent-specific measure for young people in the care system – the Assessment Checklist for Adolescents. 2. Selection and refinement of ACA content The central strategy for selecting and developing item content for the ACA was to refine and build on the content development carried out for the ACC a decade previously, rather than devising content from scratch. Prior to its inclusion in the CICS baseline survey, the ACC‟s clinical content was systematically derived using a combination of deductive and inductive strategies (Tarren-Sweeney, 2007), with the aim of identifying all clinically significant problems experienced by children and young people in alternate care that are not adequately measured by the CBCL. These strategies were (in order of occurrence): 1. a review of 110 clinical assessment reports sourced from the psychological records of 50 children and young people in care; 2. a survey of 47 clinicians (Psychiatrists and Psychologists) who worked extensively or exclusively with NSW children in care and / or maltreated children; 3. a review of literature describing the mental health of maltreated children, and of children in foster, kinship and residential care; development of a conceptual framework in the form of hypothesised symptom clusters; 4. review of a draft instrument by a foster parent focus group; 5. a final review of the proposed items by a group of experienced clinicians The present steps taken to build on and refine content for an adolescent-specific instrument are listed below: 2.1 Reviewing age-sensitive / age-appropriate ACC items The initial step was to identify two sets of age-related items from the original pool of 132 ACC items, using both age-item correlations, and clinical-developmental reasoning. First, several items were identified as behaviours that are largely manifested among younger children, and which are developmentally-based (such as pica-related behaviours and masturbation in view of others). Second, several items were identified as being maladaptive during early and middle 3 The Assessment Checklist for Adolescents - ACA childhood, but likely to be increasingly normative at ages 15 to 17 (e.g. some of the non-coercive sexual behaviours; cutting or ripping one‟s clothes). Further to this, two items measuring inattention were removed because „inattention‟ is adequately measured by the CBCL. Thirteen items were thus removed from the ACC 132-item pool. 2.2 State-wide survey of clinicians Approximately forty Psychologists employed by the Psychological service of the NSW statutory child welfare agency, who carry out clinical assessments of young people in care, were invited to propose any: 1. individual problematic behaviours, emotional states, traits, or manners of relating to other people, that are particularly observed among adolescents in care aged 12 to 17 years, and which are not adequately measured by the ACC or CBCL; and 2. mental health disorders, syndromes or constructs that are particularly manifested by adolescents in care, and which are not presently measured by the ACC or CBCL. Eight clinicians responded to the survey. Several respondents proposed that adolescents in care present with trauma-related PTSD and dissociative symptoms that are not measured by the ACC. Others emphasised the need to obtain more comprehensive measurement of emotional and behavioural dysregulation. 2.3 Literature review A review was carried out of literature describing the mental health of maltreated young people, and of adolescents in foster, kinship and residential care, updating a review carried out a decade earlier during development of the ACC. 2.4 Development of a conceptual framework The steps carried out to this point supported the retention of content contained in the ACC‟s clinical scales, as well as retention of the two low self-esteem scales, with some modifications to the item pool and item wording. From the collated information, a conceptual framework was developed in the form of hypothesised symptom clusters additional to the ACC‟s 10 clinical scales. The conceptual framework is summarised in Table 1, centring on trauma and dysregulation symptoms. A number of items measuring dissociation were included in the 132- item research version of the ACC, but only a weak dissociation factor was identified at that time. It was thought that the dissociation items written for the ACC may not adequately measure the forms of dissociation manifested by children and young people in care. Further work was done to construct improved dissociation items to pilot in the CICS follow-up and adolescent surveys. Whereas clinicians and the literature did not identify restrictive eating symptoms as being 4 The Assessment Checklist for Adolescents - ACA characteristically observed among this population, the present research presented an opportunity to examine the utility and relevance of measuring eating disorder symptoms among young people in care. 2.5 Modification of items for an adolescent population Fifteen ACC items were modified so as to have greater validity for an adolescent population. The modified ACA items are listed alongside the original ACC items, together with the reasoning for each modification, in Table 2. 2.6 Proposed new items reviewed by foster parents A list of proposed new items was reviewed by two groups of foster parents (12 foster parents in total), who were attending research focus groups in Christchurch, New Zealand, as part of a study of caregiver perceptions of children and young people‟s reactions to birth family contact. Following the completion of the focus groups, the group participants were invited to review the readability and validity of the list of proposed new ACA items within an informal discussion. Their feedback prompted changes to the wording of three of the items. 2.7 Inclusion of a draft instrument in the CICS follow-up and adolescent surveys A draft version of the ACA containing 136 items (118 clinical items, 18 low self-esteem items) was completed by foster and kinship caregiver respondents in the CICS follow-up and adolescent surveys. The draft instrument contained 111 existing or revised items from the 132- item research version of the ACC, and 25 newly developed items. [insert Tables 1 and 2 about here] 3. Checklist format The ACA administration format is identical to the ACC. Each item refers to an individual behaviour, emotional state, trait, or manner of relating to others, that is observable by a child‟s carer. There are separate versions for boys and girls, allowing for use of gender-specific personal pronouns (him/her, himself/herself, he/she). Items are otherwise identical for boys and girls. 3.1 Response scale The ACA employs a three-point response scale (0-1-2), as used by the Revised Rutter Scales (“does not apply”, “applies somewhat”, “certainly applies”) (Hogg, Rutter, & Richman, 1997), the CBCL and its companion instruments (“not true”, “somewhat or sometimes true”, “very true or often true”) (Achenbach & Rescorla, 2001), and the Strengths and Difficulties Questionnaire (“not true”, “somewhat true”, “certainly true”). 5 The Assessment Checklist for Adolescents - ACA 3.2 Measurement of infrequent critical events The ACA and ACC were designed to detect infrequent events of critical problems such as suicide attempts and discourse, age-inappropriate sexual behaviour, and self-injury. Detecting single or isolated behaviours is important if they are markers for risk of harm, or if the events have clinical significance. Conversely, it is not useful to detect isolated instances of less critical problems, such as peer conflict. The ACA and ACC differentiate between these two types of items by assigning them to separate parts, each of which employs a different three-step response scale. Part 1 uses the following instructions for less critical / higher incidence problems: “Circle 0 if the statement is not true for this young person, in the last 4 to 6 months” “Circle 1 if the statement is partly true for this young person, in the last 4 to 6 months” “Circle 2 if the statement is mostly true for this young person, in the last 4 to 6 months” Part 2 uses the following instructions for more critical / lower incidence problems: “Circle 0 if the behaviour did not occur in the last 4 to 6 months” “Circle 1 if the behaviour occurred once in the last 4 to 6 months” “Circle 2 if the behaviour occurred more than once in the last 4 to 6 months” 4. Item and factor analyses 4.1 Sample Item and factor analyses were performed on ACA scores for 372 young people in long- term alternate care, obtained from the CICS follow-up survey (n=85) and the CICS adolescent survey (n=147). Some analyses were supplemented by de-identified ACC scores for 142 young people enrolled in the multi-agency treatment foster care programme of the Children's Aid Societies of Durham, Highland Shores and Kawartha-Haliburton, in Ontario, Canada. The CICS is a prospective, epidemiological study of the mental health and development of children and young people in court-ordered foster and kinship care, in NSW, Australia. The follow-up and adolescent survey stages of the CICS have human ethics approvals from the University of Newcastle in Australia, and the University of Canterbury in New Zealand. The CICS was also approved and funded by the NSW Department of Family and Community Services (the state child welfare agency). Otherwise the Department had no role in the analysis or dissemination of the CICS findings. The CICS follow-up study attempted to obtain follow-up social care and developmental data for all participants in the CICS baseline survey (carried out between 1999 and 2000) who remained in foster or kinship care in early to mid 2009. Of the 347 baseline survey participants, 6 The Assessment Checklist for Adolescents - ACA 231 remained in court-ordered foster or kinship care. Of these, 70 were residing in placements that did not have verifiable contact details, and whose caregivers could not be located by telephone. Of the remaining 171 eligible participants, questionnaires were returned for 85 young people, representing a 50% response rate. However, these participants represent only 25% of the baseline sample. The CICS adolescent survey was a cross-sectional survey of the mental health of 11 to 17 year-olds in long-term, court-ordered foster and kinship care in NSW, Australia, carried out a year after the CICS follow-up survey. The sampling frame was all young people aged 11 to 17 years residing in non-temporary court-ordered foster and kinship care in New South Wales, Australia, where case supervision was provided by the statutory child welfare agency (i.e. not supervised by private agencies), who were not part of the CICS baseline survey sample (i.e. baseline survey participants and non-participants), and whose placement address could be verified. Survey questionnaires were sent to the caregivers of 290 eligible participants residing at residential addresses with telephone contact, of which 145 were completed and returned (50% response rate). The young people ranged in age from 11 to 18 years, with a mean age of 14.8 years. There was uneven distribution of young people across the age ranges, with fewer numbers in the oldest and youngest age groups (see Table 3). The mean age of young people in the three samples outlined above was 15.6, 15.2 and 14.0 years respectively. Gender was also unevenly distributed (54% boys and 46% girls). Eighty-three percent of young people resided with foster parents, 13% with kin, and 4% with adoptive parents. Their mean age at entry into care was 6.2 years and mean time in care (including post-care adoption) was 8.6 years. A total of 78% of young people had clinic referred status, as defined by one or more of the following criteria: 1. residing in treatment foster care; 2. prescribed psychopharmacologic medication; 3. caregiver receiving behavioural support or other carer intervention or young person receiving individual or group psychotherapy or counselling; and 4. caregiver actively seeking referral to a mental health service. [insert Table 3 about here] 4.2 Item analysis A nominal total clinical score was generated from the sum of clinical item scores, and item-total correlations were calculated for the combined CICS samples (n=230). All but one of the item-total correlations were positive (the exception was excessive dieting or fasting), with 84% exceeding r = .30, and 40% exceeding r = .50. Eight items correlated less than r = .20. The procedure was repeated for low self-esteem items. The lowest low self-esteem item-total correlation was 0.44, with all others exceeding r = .50. 7 The Assessment Checklist for Adolescents - ACA Excluding items that have an aggregate mean score less than .2, there were 19 items for which the gender ratio of item mean scores exceeded 1.5 (boy > girl = 11 items; girl > boy = 8 items). Sixteen items had an item-age correlation between .10 and .20 (9 positive and 7 negative), and there was one item-age correlation higher than .20 (uses drugs or alcohol other than prescribed medications, r = .31). The latter correlation possibly reflects the increasingly normative use of alcohol in late adolescence. The analyses revealed considerably less age variation in item scores across the 11 to 18 year age-span, than was found across the 4 to 11 year age-span during the development of the ACC, possibly because the present data appear to be less confounded by age at entry into care. Clinical item prevalence, defined as the proportion of the sample scoring either 1 or 2, ranged from 0.8% to 69%. Seven items had greater than 50% prevalence, such that they might be considered to be characteristic difficulties of young people in care. Five of these were considered to be sufficiently maladaptive to retain for further analysis, while attention-seeking behaviour, and easily influenced by other young people were not retained. Another seven items had very low prevalence and were not retained for factor analysis, including several that refer to restrictive eating. An initial exploratory factor analysis included these items but did not locate a restrictive eating factor. Consequently, a total of nine clinical items were discarded prior to factor analysis. 4.3 Factor analysis of clinical items A series of principal components factor analyses were carried out using oblique (promax) rotations. The analyses were hampered by a relatively low sample size for the number of items under examination. To work around this limitation, a number of exploratory factor analyses were first carried out to: 1. identify those factors that remain robust when selecting different sets of items; and 2. identify those items that do not load on any clinically meaningful factors through various rotations. Those items that performed weakly were progressively removed from the analyses, with a view to maximizing the subject to item (STI) ratio. Ten factors were initially retained after performing a scree plot test, each of which had an eigenvalue greater than 1.5. Rotations were then performed on models of 5 to 10 factors. The various analyses suggested a 7- factor model yielded the most stable and meaningful factors, while an 8-factor model yielded several additional weak factors, depending on which items were retained for analysis. Within the 7-factor model, two factors described very discrete constructs (food maintenance behaviour, and sexual behaviour problems). These factors remained robust regardless of which items were included in the analyses. This made it possible to carry out separate exploratory factor analyses of those items loading on the other five factors, again with a view to increasing the STI ratio. The exploratory analyses identified a set of items that reliably loaded .40 or higher on at least one factor. 8 The Assessment Checklist for Adolescents - ACA Two main factor analyses were then carried out. The first identified loadings on the food maintenance and sexual behaviour factors in a 7-factor model using a pool of 72 items, with an STI ratio of 3.2. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy for these 72 items was .83 overall, confirming that the item pool was appropriate for factor analysis. This model accounted for 51% of the score variance. The second identified loadings for the remaining five factors using a pool of 58 items (which excluded the food maintenance and sexual behaviour items), providing an STI ratio of 3.9, and with a KMO value of .85. This 5-factor model accounted for 47% of the score variance. 4.4 Factor analysis of low self-esteem items A preliminary factor analysis examined the factor structure of a combined pool of low self-esteem items and the 59 clinical items that loaded on the 5-factor model described above. All of the low self-esteem items loaded on two low self-esteem factors within a 7-factor model, while none of the clinical items loaded on the self-esteem factors. This suggests that the low self- esteem and clinical items are both conceptually and statistically distinct. A principal components factor analysis with oblique rotations was then performed on the 18 low self-esteem items. This yielded the same two-factor model of low self-esteem identified among younger children during the development of the ACC, representing constructs of negative self-image and low confidence. 5. Scale construction Clinical and low self-esteem scales were derived from items with factor loadings above .40 on the relevant 7-factor and 5-factor clinical models, and the 2-factor low self-esteem model. Surprisingly, no item loaded above .40 on more than one factor. An additional criterion was that an item‟s presence should not lower the internal consistency of a scale. Seventy two items met criteria for inclusion with a clinical scale. Fifteen items that were considered clinically important but did not meet criteria for inclusion were retained as other items. One such item (does not show pain if physically hurt) loaded .44 on a dissociation factor, but its presence would have lowered that scale‟s internal consistency. All of the low self-esteem items met inclusion criteria. The final version of the ACA thus contains 105 items (87 clinical items and 18 low self-esteem items). All of the low self-esteem items and 66 of the 87 clinical items are shared with the ACC. Four of the ACA clinical scales have equivalents on the ACC, while the remaining three describe constructs that are not measured by the ACC. The ACA scales are listed with their constituent items in the Appendix. 9 The Assessment Checklist for Adolescents - ACA 6. Scale properties The clinical scales are labelled: I. Non-reciprocal interpersonal behaviour II. Social instability / behavioural dysregulation * III. Emotional dysregulation / distorted social cognition * IV. Dissociation / trauma symptoms * V. Food maintenance VI. Sexual behaviour VII. Suicide discourse The low self-esteem scales are labelled: I. Negative self-image II. Low confidence * Scale is unique to the ACA 6.1 Relationships between clinical scales A correlation matrix of the ACA clinical scales is presented in Table 4. The clinical scales are numbered I to VII in the order that best reflects inter-scale correlations and symptom prevalence i.e. adjacent scales are more closely related in terms of both correlation and the commonality of the difficulties being measured. A principal components factor analysis conducted on the seven clinical scale scores indicates the instrument has no higher-order factor structure (such as internalizing versus externalizing factors). [insert Table 4 about here] 6.2 Optimal cut-points for ACA total clinical score Relationships between ACA total score distributions and various indicators of mental health impairment were examined across the three study samples, with a view to identifying clinically significant scores. The indicators were: having CBCL total problems scores in the clinical range, and „borderline plus clinical‟ ranges (CICS); having a reported psychiatric diagnosis (CICS); having prescribed psychopharmacologic medication(s) (CICS); and having ACC total scores in the clinical and „elevated plus clinical‟ ranges (Canadian TFC sample). Sensitivity and specificity were plotted for each criterion in Receiver Operating Characteristics (ROC) analyses. An instrument‟s screening accuracy is measured by the area under the ROC curve, referred to as the AUC, which is expressed as a proportion. An instrument that predicts no better than chance will have an AUC value in the vicinity of 0.5, while a „perfect‟ screening instrument has an AUC of 1.0. Unless an instrument has a score cut-point that is 100% accurate (i.e. incurs no false positive or false negative results), then every screening cut-point involves a trade-off between sensitivity (the proportion of clinical cases who are positively screened i.e. 10

Description:
(ACA), which is an adolescent version of the Assessment Checklist for Children ( ACC). The. ACC is a 120-item carer-report psychiatric rating instrument, that
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.